nixpkgs/pkgs/tools/networking/linkchecker/add-no-robots-flag.patch
Peter Hoeg c1ffbd8ee8 linkchecker: init at 9.3
A few changes from upstream:

 1) the executable is patched to support a --no-robots flag to ignore
    robots.txto

 2) the GUI doesn't work (for now), so this is CLI only
2016-05-27 15:07:05 +08:00

61 lines
2.7 KiB
Diff

diff --git a/linkcheck/checker/httpurl.py b/linkcheck/checker/httpurl.py
index 6f207b6..161619c 100644
--- a/linkcheck/checker/httpurl.py
+++ b/linkcheck/checker/httpurl.py
@@ -75,7 +75,7 @@ def allows_robots (self, url):
@return: True if access is granted, otherwise False
@rtype: bool
"""
- return self.aggregate.robots_txt.allows_url(self)
+ return not self.aggregate.config['robotstxt'] or self.aggregate.robots_txt.allows_url(self)
def content_allows_robots (self):
"""
diff --git a/linkcheck/configuration/__init__.py b/linkcheck/configuration/__init__.py
index fc2c148..234fa05 100644
--- a/linkcheck/configuration/__init__.py
+++ b/linkcheck/configuration/__init__.py
@@ -163,6 +163,7 @@ def __init__ (self):
## checking options
self["allowedschemes"] = []
self['cookiefile'] = None
+ self['robotstxt'] = True
self["debugmemory"] = False
self["localwebroot"] = None
self["maxfilesizeparse"] = 1*1024*1024
diff --git a/linkcheck/configuration/confparse.py b/linkcheck/configuration/confparse.py
index 67751ed..845fa95 100644
--- a/linkcheck/configuration/confparse.py
+++ b/linkcheck/configuration/confparse.py
@@ -149,6 +149,7 @@ def read_checking_config (self):
self.get(section, 'allowedschemes').split(',')]
self.read_boolean_option(section, "debugmemory")
self.read_string_option(section, "cookiefile")
+ self.read_boolean_option(section, "robotstxt")
self.read_string_option(section, "localwebroot")
try:
self.read_boolean_option(section, "sslverify")
diff --git a/linkchecker b/linkchecker
index 199532c..9e91fa5 100755
--- a/linkchecker
+++ b/linkchecker
@@ -321,6 +321,9 @@ group.add_argument("--cookiefile", dest="cookiefile", metavar="FILENAME",
help=_(
"""Read a file with initial cookie data. The cookie data format is
explained below."""))
+# const because store_false doesn't detect absent flags
+group.add_argument("--no-robots", action="store_const", const=False,
+ dest="norobotstxt", help=_("Disable robots.txt checks"))
group.add_argument("--check-extern", action="store_true",
dest="checkextern", help=_("""Check external URLs."""))
group.add_argument("--ignore-url", action="append", metavar="REGEX",
@@ -431,6 +434,8 @@ if options.externstrict:
if options.extern:
pats = [linkcheck.get_link_pat(arg) for arg in options.extern]
config["externlinks"].extend(pats)
+if options.norobotstxt is not None:
+ config['robotstxt'] = options.norobotstxt
if options.checkextern:
config["checkextern"] = True
elif not config["checkextern"]: